本篇論文欲探討影響台灣股市系統性風險係數檢定結果之原因。歸納原因有四，分別為投組報酬率計算方式不同、指數選取的不同、樣本選取的不同以及分組方式的不同。實證結果顯示，在其他條件不變下，僅改變投資組合報酬率的計算方式，不論以加權平均法或者簡單平均法計算投資組合報酬率，對系統性風險係數的檢定結果並無影響。在其他條件不變下，僅改變指數的選取，不論以加權股價指數代理市場指數或者以報酬指數代理市場指數，對系統性風險係數的檢定結果並無影響。 在其他條件不變下，僅改變樣本的選取，不論樣本為上市公司，包含金融類股但排除全額交割股，或者樣本為上市、上櫃公司，包含金融類股以及全額交割股為下市前的資料，對系統性風險係數的檢定結果並無影響。 當其他條件不變下，改變分組方式，會強烈地影響系統性風險係數的檢定結果。由於二階段橫斷面迴歸分析中，必須以組beta估計值替代個股beta值，又分組方式的不同會影響組beta的估計值，進而影響第二階段的橫斷面迴歸分析結果。 This thesis wants to examine what affect the results of the test of beta coefficient. We sum up four reasons: first, the method of calculating the returns of the portfolio. Second, the difference of selecting the market index. Third, the difference of selecting the sample. Last, the difference of grouping. The empirical results reveal that the method of calculating the returns of the portfolio will not affect the results of the test of beta coefficient. The empirical results reveal that the difference of selecting the market index and the sample will not affect the results of the test of beta coefficient, neither. We find that only the the difference of grouping will affect the results of the test of beta coefficient. Because the difference of grouping will affect the estimation of group beta and in the two-step cross section regresssion analysis we have to put the group beta into the individual stock, the difference of grouping will affect the results of the test of beta coefficient.
